摘要
针对传统的亚像素边缘检测技术需要先进行粗定位,然后再进行精确定位的问题,提出无粗定位亚像素边缘检测算法.针对彩色图像降低维度边缘信息损失的问题,提出主轴分析法来保存图像的边缘信息.针对滤波后图像边缘特征模糊的问题,提出基于二分K均值的中值滤波算法,在滤除噪声的同时,增强图像的边缘.最后从检测精度、检测效率和可靠性3个方面验证算法的有效性.
Concerning the traditional sub-pixel edge detection techniques need to use the traditional edge detection algorithms for coarse positioning before precision positioning the algorithm of sub-pixel edge detection without rough localization was proposed. Concerning the edge information loss of color image principal axis analysis method was proposed to preserve the edge information. Concerning image edge feature gets blurred after filtering a median filter algorithm based on bisecting K-means was proposed. Bisecting K-means median filter algorithm can enhance the image edges when filtering out the noise. At the end by comparing the result at the aspects of detection accuracy, efficiency and reliability the validity of the algorithm proposed can be proved.
出处
《上海师范大学学报（自然科学版）》
2016年第4期434-440,共7页
Journal of Shanghai Normal University(Natural Sciences)
关键词
边缘检测
亚像素
粗定位
主轴分析法
二分K均值
edge detection
sub-pixel
rough localization
principal axis analysis method
bisecting K-means
